Skip to main content
Figure 5 | EPJ Data Science

Figure 5

From: Multilayer networks for text analysis with multiple data types

Figure 5

Text layer determines the partitions obtained in the multilayer model (Hyperlink + Text). Similarity (overlap of consensus partition) between Hyperlink partition and μ-Hyperlink-Text partition as a function of the subsampling parameter μ where \(\mu =0\) (\(\mu =1\)) corresponds to the case with all (none) of the word tokens removed in the Hyperlink-Text model. For a given value of μ, a random fraction \(1-\mu \) of the words were removed and the Hyperlink + Text model was then fitted for multiple iterations. The consensus partition was then computed for the Hyperlink + Text model and its partition overlap with Hyperlink model. A higher sub-sampling of text (i.e. smaller values of μ) results in the consensus partition between the Hyperlink + Text and Hyperlink model having a high degree of overlap

Back to article page